Pith. sign in

REVIEW 2 cited by

Multiplayer bandits without observing collision information

Not yet reviewed by Pith; the record is open.

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 1808.08416 v2 pith:PU37QSLD submitted 2018-08-25 cs.LG cs.GTstat.ML

classification cs.LGcs.GTstat.ML
keywords collisionplayersalgorithmfirstgivemodelregretdepend
verification ladder T0 review T1 audit T2 compute T3 formal

Signed reviews

No signed human review yet.

0 comments
read the original abstract

We study multiplayer stochastic multi-armed bandit problems in which the players cannot communicate and if two or more players pull the same arm, a collision occurs and the involved players receive zero reward. We consider two feedback models: a model in which the players can observe whether a collision has occurred and a more difficult setup when no collision information is available. We give the first theoretical guarantees for the second model: an algorithm with a logarithmic regret, and an algorithm with a square-root regret type that does not depend on the gaps between the means. For the first model, we give the first square-root regret bounds that do not depend on the gaps. Building on these ideas, we also give an algorithm for reaching approximate Nash equilibria quickly in stochastic anti-coordination games.

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 2 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Multiplayer Bandit Learning, from Competition to Cooperation

    cs.GT 2019-08 conditional novelty 6.0 of 10

    In a two-player bandit game where players see each other's actions but not rewards, competition reduces exploration, cooperation increases it, and neutral players can outperform a single player by observing each other.

  2. Accelerated learning from recommender systems using multi-armed bandit

    cs.IR 2019-08 conditional novelty 4.0 of 10

    A Vrbo team used daily Thompson sampling to rank four recommendation models by click-through rate, but the A/B validation they report is for a previous campaign's winner, not the current one.

Pith tools